Abstract: Text filtering plays an important role in processing readable online texts. A hybrid model for text filtering is presented in this paper. Its main idea is shown as follows: it integrates content-based filtering and collaborative filtering. On the basis of users’ annotations, it provides users with the texts related to their interest. Also put forward are the approach for text feature extraction, the pattern of user annotations, and the mechanism for matching texts and profiles. As a result, the model is independent of concrete domain knowledge, and it improves the efficiency of text filtering.